Shengchao Liu

I am a Tenure-Track Assistant Professor in Computer Science and Engineering at The Chinese University of Hong Kong, and the Principal Investigator of the Wave Intelligence Lab. His research lies at the intersection of artificial intelligence and physics, with two main focuses: (1) developing physics-inspired AI methods to accelerate existing scientific paradigms, as well as leveraging generative models to enable new paradigms of scientific discovery, with applications in chemistry, materials science, biology, and geoscience; and (2) using principles from physics to understand and characterize the internal mechanisms of AI models, including emerging paradigms of generative modeling, latent space representations, and phase transitions in optimization and training dynamics.

  • E-Mail: shengchao1224 at gmail dot com
  • Highlights:
    I want to share some inspiring research values:
    • 科研的意义，不在于起点，而在于终点。
    • 科研的核心秘诀在于提出正确的问题 —— 如果一个问题过于困难，也许并不是我们能力不够，而是问题本身还没有被问对。
      • (Special thanks to Weiyang for the following reflections)
    • Focus on creating novel ideas, not publishing papers
    • Follow curiosity and passion, not trends
    • Ideas are not owned, but come with debts to those who came before
    • Ideas become stronger when shared, discussed and criticized
    • Life is surprisingly short, so solve problems that interest and excite you most
    • People who wish to analyze nature without using mathematics must settle for a reduced understanding. -- Richard P. Feynman
    • It is good to be quick, but it is more important to be deep
    • Think like an amateur, do as an expert
    • Last lecture by Prof. Randy Pausch
    For prospective students:
    • Due to the high volume of emails, I apologize if I haven’t been able to reply individually.
    • I am always looking for self-motivated PhD students, Postdoc, and RAs/visitors/interns.
    • Solid math/engineering and good communication skills are necessary.
    • We have posts introducing our lab on Zhihu and Xiaohongshu (RedNote). Please feel free to take a look.
    • There is NO need to email me to apply. Simply fill out this form, and I will follow up within two weeks. Thank you for your interest!
    • If you’re interested in AI but don’t know where to start, please refer to this tutorial and this reading list。


Research Interest

  • AI and DL
    • Representation Learning: structured / geometric / graph representation learning
    • Transfer Learning: self-supervised learning, multi-task learning
  • Science for AI
    • Deep Generative Modeling (GenAI): LLM, EBM, controllable / conditional generative modeling
    • Learning Dynamics: optimization, training dynamics, architecture design
    • Latent Space Interpolation: representation, semantics, reasoning
  • AI for Science
    • Molecule Representation and Inverse Design: small molecules, proteins, crystal materials, genomics
    • Physical Dynamics: classical molecular dynamics, ab initio molecular dynamics


Selected Publications

Notice: Please click a tab to view the corresponding articles.
  • A Minimal Model of Representation Collapse: Frustration, Stop-Gradient, and Dynamics
    Louie Hong Yao*, Yuhao Li*, Shengchao Liu
    [ArXiv]
  • InertialAR: Autoregressive 3D Molecule Generation with Inertial Frames
    Haorui Li, Weitao Du, Yuqiang Li, Hongyu Guo, Shengchao Liu
    ICML 2026
    [ArXiv] [Code]
  • ChatBattery: Expert-Guided LLM Reasoning for Battery Discovery: From AI-Driven Hypothesis to Synthesis and Characterization
    Shengchao Liu*, Hannan Xu*, Yan Ai*, Huanxin Li*, Yoshua Bengio*, Hongyu Guo*
    [ArXiv] [Code]
    [Featured in The Globe and Mail]
  • RigidSSL: Rigidity-Aware Geometric Pretraining for Protein Design and Conformational Ensembles
    Zhanghan Ni*, Yanjing Li*, Zeju Qiu*, Bernhard Schölkopf, Hongyu Guo, Weiyang Liu, Shengchao Liu
    ICLR 2026
    [Paper] [ArXiv] [BioRxiv] [Code]
  • InertialGenome: A Resolution-Agnostic Geometric Transformer for Chromosome Modeling Using Inertial Frame
    Yize Zhou, Haorui Li, Shengchao Liu
    ICLR 2026
    [Paper] [BioRXiv] [Code]
  • K-Flow: Flow Along the K-Amplitude for Generative Modeling
    Weitao Du, Shuning Chang, Jiasheng Tang, Yu Rong, Fan Wang, Shengchao Liu
    ICLR 2026
    [Paper] [ArXiv] [Code]
    [ICLR Deep Generative Model Workshop 2025 Oral]
  • NeuralMD: A Multi-Grained Symmetric Differential Equation Model for Learning Protein-Ligand Binding Dynamics
    Shengchao Liu*, Weitao Du*, Hannan Xu, Yanjing Li, Zhuoxinran Li, Vignesh Bhethanabotla, Divin Yan, Christian Borgs*, Anima Anandkumar*, Hongyu Guo*, Jennifer Chayes*
    Nature Communications 2025
    [Paper] [Arxiv] [Project Page] [Code]
    [ICLR AI4DifferentialEquations Workshop 2024 Oral]
  • AssembleFlow: Rigid Flow Matching with Inertial Frames for Molecular Assembly
    Hongyu Guo, Yoshua Bengio, Shengchao Liu
    ICLR 2025
    [Paper] [Code]
  • NucleusDiff: Manifold-Constrained Nucleus-Level Denoising Diffusion Model for Structure-Based Drug Design
    Shengchao Liu*, Divin Yan*, Weitao Du, Weiyang Liu, Zhuoxinran Li, Hongyu Guo, Christian Borgs*, Jennifer Chayes*, Anima Anandkumar*
    Proceedings of the National Academy of Sciences (PNAS) 2025
    [Paper] [ArXiv] [Project Page]
    [ICML GRaM Workshop 2024] [Caltech News]
  • NeuralCrystal: A Geometric Foundation Model for Crystalline Material Discovery
    Shengchao Liu*, Divin Yan*, Weitao Du, Zhuoxinran Li, Zhiling Zheng, Omar Yaghi, Christian Borgs, Hongyu Guo, Anima Anandkumar, Jennifer Chayes
    [NeurIPS AI4Mat Workshop 2024]
  • CrystalFlow: An Equivariant Flow Matching Framework for Learning Molecular Crystallization
    Shengchao Liu, Divin Yan, Hongyu Guo*, Anima Anandkumar*
    [ICML ML4LMS Workshop 2024] [ICML GRaM Workshop 2024]
  • ChatDrug: ChatGPT-powered Conversational Drug Editing Using Retrieval and Domain Feedback
    Shengchao Liu*, Jiongxiao Wang*, Yijin Yang, Chengpeng Wang, Ling Liu, Hongyu Guo, Chaowei Xiao
    ICLR 2024
    [Paper] [Arxiv] [Project Page] [Code]
    [ICML SynS and ML Workshop 2023 Oral]
  • ChatPathway: Conversational Large Language Models for Biology Pathway Detection
    In Preparation
    First version in [NeurIPS GLFrontiers Workshop 2023 Oral]
  • ProteinDT: A Text-guided Protein Design Framework
    Shengchao Liu, Yanjing Li, Zhuoxinran Li, Anthony Gitter, Yutao Zhu, Jiarui Lu, Zhao Xu, Weili Nie, Arvind Ramanathan, Chaowei Xiao*, Jian Tang*, Hongyu Guo*, Anima Anandkumar*
    Nature Machine Intelligence 2025
    [Paper] [Arxiv] [Project Page] [Code]
  • MoleculeSTM: Multi-modal Molecule Structure-text Model for Text-based Editing and Retrieval
    Shengchao Liu, Weili Nie, Chengpeng Wang, Jiarui Lu, Zhuoran Qiao, Ling Liu, Jian Tang, Chaowei Xiao, Anima Anandkumar
    Nature Machine Intelligence 2023
    [Paper] [Arxiv] [Project Page] [Code]
    [NeurIPS AI4Science Workshop 2022]
  • GraphCG: Unsupervised Discovery of Steerable Factors When Graph Deep Generative Models Are Entangled
    Shengchao Liu, Chengpeng Wang, Jiarui Lu, Weili Nie, Hanchen Wang, Zuoxinran Li, Bolei Zhou, Jian Tang
    TMLR 2024
    [Paper] [Arxiv] [Project Page] [Code]
    [NeurIPS GLFrontiers Workshop 2022 Oral]
  • Geom3D: Symmetry-Informed Geometric Representation for Molecules, Proteins, and Crystalline Materials
    Shengchao Liu, Weitao Du, Yanjing Li, Zhuoxinran Li, Zhiling Zheng, Chenru Duan, Zhiming Ma, Omar Yaghi, Anima Anandkumar, Christian Borgs, Jennifer Chayes, Hongyu Guo, Jian Tang
    NeurIPS Datasets and Benchmarks 2023
    [Paper] [Arxiv] [Code]
  • SpaTea: A Quantum-Inspired Neural Network for Geometric Modeling
    Weitao Du*, Shengchao Liu*, Hongyu Guo
    In Preparation
    [Arxiv]
  • MoleculeJAE: Molecule Joint Auto-Encoding: Trajectory Pretraining with 2D and 3D Diffusion
    Weitao Du, Jiujiu Chen, Xuecang Zhang, Zhiming Ma, Shengchao Liu
    NeurIPS 2023
    [Paper] [Arxiv] [Code]
  • MoleculeSDE: A Group Symmetric Stochastic Differential Equation Model for Molecule Multi-modal Pretraining
    Shengchao Liu*, Weitao Du*, Zhiming Ma, Hongyu Guo, Jian Tang
    ICML 2023
    [Paper] [Arxiv] [Project Page] [Code]
  • GeoSSL: Molecular Geometry Pretraining with SE(3)-Invariant Denoising Distance Matching
    Shengchao Liu, Hongyu Guo, Jian Tang
    ICLR 2023
    [Paper] [Arxiv] [Project Page] [Code]
  • GraphMVP: Pre-training Molecular Graph Representation with 3D Geometry
    Shengchao Liu, Hanchen Wang, Weiyang Liu, Joan Lasenby, Hongyu Guo, Jian Tang
    ICLR 2022
    [Paper] [Arxiv] [Project Page] [Code]
    [NeurIPS SSL Workshop 2021]
    [ICLR GTRL Workshop 2022 Spotlight]
  • SGNN-EBM: Structured Multi-task Learning for Molecular Property Prediction
    Shengchao Liu, Meng Qu, Zuobai Zhang, Huiyu Cai, Jian Tang
    AISTATS 2022
    [Paper] [Arxiv] [Project Page] [Code]
    [NeurIPS AI4Science Workshop 2021]
  • AWARE: Attentive Walk-Aggregating Graph Neural Networks
    Mehmet F. Demirel, Shengchao Liu, Siddhant Garg, Zhenmei Shi, Yingyu Liang
    TMLR 2022
    [Paper] [Arxiv] [Code]
  • LBTW: Loss-Balanced Task Weighting to Reduce Negative Transfer in Multi-Task Learning
    Shengchao Liu, Yingyu Liang, Anthony Gitter
    AAAI-Student Abstract 2019
    [Paper] [Appendix] [Code]
  • Bad Global Minima Exist and SGD Can Reach Them
    Shengchao Liu, Dimitris Papailiopoulos, Dimitris Achlioptas
    NeurIPS 2020
    [Paper] [Code] [Video, NeurIPS 2020]
    [ICML Deep Learning Phenomena Workshop 2019 Oral]
  • N-Gram Graph: Simple Unsupervised Representation for Graphs, with Applications to Molecules
    Shengchao Liu, Mehmet Furkan Demirel, Yingyu Liang
    NeurIPS 2019 Spotlight
    [Paper] [Arxiv] [Code]
    [NeurIPS MLMM Workshop 2018]


Previously ...

I was a postdoctoral researcher at UC Berkeley, where I worked with Prof. Jennifer Chayes and Prof. Christian Borgs. I received my Ph.D. in Computer Science from the Quebec Artificial Intelligence Institute (Mila) and Université de Montréal under the supervision of Prof. Jian Tang. I earned my Master’s degree in Computer Science from the University of Wisconsin–Madison, where I was also a graduate researcher at the Morgridge Institute for Research, and initiated my first research project under the guidance of Prof. Anthony Gitter, Prof. Yingyu Liang, and Prof. Dimitris Papailiopoulos. Prior to that, I obtained my Bachelor’s degree from Shandong University. I have also gained industry experience through internships at Facebook, IQVIA, ServiceNow AI, and NVIDIA.